Shifting Immigration Toward High-Skilled Workers
Shifting Immigration Toward High-Skilled Workers
We evaluate two immigration policies that shift 10 percent of future low-skilled immigration toward either: (i) high-skilled immigrants (“ HSI ”) that otherwise maintains the current share of STEM workers within the high-skilled group, or (ii) only high-skilled STEM workers (“ HSI STEM ”) that increases the share of STEM relative to other high-skill workers. The number of total immigrants remains the same under both policies. Both policies grow the economy, reduce federal debt, and increase wages across all income groups: lower-skilled, higher-skilled non-STEM workers, and higher-skilled STEM workers. In fact, this policy change affords the rare opportunity of a “Pareto improvement” benefitting all groups.
Key Points
-
10 percent of low-skilled immigrants equals just 25,000 workers in 2025, or about 0.01 percent (one percent of one percent) of the total U.S. working-age population. Over the next 30 years, 750,000 immigrant workers would be converted from low to high skill, still representing just 0.35 percent of the total U.S. working-age population in 2054. The total number of immigrant workers remains unchanged.
-
This policy has an outsized impact on economic growth relative to the size of the policy change. We project that the HSI policy would increase GDP by 0.1 percent by 2034 and 0.4 percent by 2054, while HSI STEM policy would increase GDP by 0.3 percent by 2034 and 0.7 percent by 2054.
-
Both policies reduce outlays and increase revenues over the 10-year budget window ending in 2034, decreasing primary deficits by $65.3 billion under the HSI policy and $152.6 billion under the HSI STEM policy.
-
All education, occupation, income, and age groups gain from both policy changes. The gains also benefit low-income workers the most relative to their earnings before these policy changes. Wages for low-skill workers increase by 0.5 percent within a decade and by 1.3 percent over the long run. Future low-skill workers born today gain between $10,000 and $16,000 in lifetime value from HSI and HSI STEM, respectively.
Under current law, total immigration is projected to increase the U.S. adult population (ages 21 and older) by 15.3 million by 2054, accounting for approximately 40 percent of growth in total adult population (see Figure 1). Based on historical trends, about 35 percent of new immigrants are expected to have a high school diploma or less, classifying them as low-skilled labor. In contrast, about 65 percent are anticipated to have some college education or higher, placing them in the high-skilled labor category. Among these high-skilled immigrants, we estimate that only about 15 percent will work in STEM occupations.
Source: Penn Wharton Budget Model
We analyze two illustrative policy changes that alter the skill composition of new immigrants while keeping the total number of projected immigrants constant. These policy changes reduce the number of low-skilled visas by 10 percent and increase the number of high-skilled visas by the same amount. The difference between the two policies is how we allocate high-skilled visas.
-
High-skill immigration (HSI): Additional visas are issued to high-skilled non-STEM and high-skilled STEM immigrants so that their proportions among all high-skilled immigrants remain consistent with current immigration projections.
-
High-skill STEM immigration (HSI STEM): All additional visas are issued to high-skill STEM immigrants.
Table 1 shows the projected number of new immigrant households entering the U.S. under current law and two illustrative policy scenarios. Between 2025 and 2054, an estimated 7.5 million low-skilled immigrants, 12.5 million high-skilled non-STEM immigrants, and 2.1 million high-skilled STEM immigrants are expected to enter the country legally through existing visa programs. By 2054, new low-skilled, high-skilled non-STEM, and high-skilled STEM immigrants will account for approximately 3.50 percent, 5.85 percent, and 1.00 percent of the total workforce, respectively.
Under the HSI and HSI STEM policies, the annual inflow of low-skilled immigrants is reduced by 10 percent, leading to a total decline of 0.75 million over the 2025 to 2054 period. By 2054, new low-skilled immigrants as a share of the workforce will decrease from 3.50 percent to 3.15 percent.
Under the HSI policy, the reduced number of low-skilled immigrants is reallocated proportionally according to the current-law distribution: 85 percent to high-skilled non-STEM and 15 percent to high-skilled STEM immigrants. This reallocation results in approximately 650,000 additional high-skilled non-STEM immigrants and 100,000 additional high-skilled STEM immigrants compared to current law. By 2054, these changes increase the new high-skilled non-STEM and new high-skilled STEM immigrants as a share of the total workforce from 5.85 percent to 6.15 percent and 1.0 percent to 1.05 percent, respectively.
Under the HSI STEM policy, the number of high-skilled non-STEM immigrants will remain unchanged from current law. In contrast, the reduction in low-skilled immigrants will be entirely offset by an increase of 750,000 high-skilled STEM immigrants. As a result, from 2025 to 2054, 2.9 million high-skilled STEM immigrants will enter the U.S., representing a 35 percent increase compared to current law. By 2054, new high-skilled STEM immigrants as a share of the workforce will increase from 1.0 percent to 1.35 percent, while the share of new high-skilled non-STEM immigrants remains unchanged at 5.85 percent.
| 2025 | 2029 | 2034 | 2039 | 2044 | 2049 | 2054 | |
|---|---|---|---|---|---|---|---|
| Current Law | |||||||
| Low-skilled | 0.3 | 1.2 | 2.4 | 3.6 | 4.9 | 6.2 | 7.5 |
| High-Skilled non-STEM | 0.4 | 2.0 | 4.1 | 6.1 | 8.2 | 10.4 | 12.5 |
| High-Skilled STEM | 0.1 | 0.3 | 0.7 | 1.0 | 1.4 | 1.8 | 2.1 |
| High-Skill Immigration | |||||||
| Low-skilled | 0.2 | 1.1 | 2.1 | 3.3 | 4.5 | 5.6 | 6.7 |
| High-Skilled non-STEM | 0.4 | 2.1 | 4.3 | 6.5 | 8.6 | 10.9 | 13.2 |
| High-Skilled STEM | 0.1 | 0.4 | 0.7 | 1.1 | 1.5 | 1.8 | 2.2 |
| High-Skill STEM Immigration | |||||||
| Low-skilled | 0.2 | 1.1 | 2.1 | 3.3 | 4.5 | 5.6 | 6.7 |
| High-Skilled non-STEM | 0.4 | 2.0 | 4.1 | 6.1 | 8.2 | 10.4 | 12.5 |
| High-Skilled STEM | 0.1 | 0.5 | 0.9 | 1.4 | 1.9 | 2.4 | 2.9 |
Note: Numbers are rounded to the nearest 100,000.
Source: Penn Wharton Budget Model
Table 2 illustrates the economic and debt impacts of the HSI and HSI STEM policies. Under both policies, GDP, capital, wages, and consumption steadily rise, while debt declines over the 2025–2054 period as more high-skilled non-STEM and STEM workers immigrate to the U.S. By 2054, under the HSI policy, GDP and capital increase by 0.4 percent, average wages rise by 0.2 percent, and debt is 0.4 percent lower than under current law (see Table 2.A). Wage growth also drives a 0.4 percent increase in consumption.
The effects are even more pronounced under the HSI STEM policy, which results in a larger influx of STEM workers. By 2054, GDP and capital will increase by 0.7 percent and 1.0 percent, respectively; average wages will be 0.5 percent higher, and debt will be reduced by 0.8 percent compared to current law (see Table 2.B). The stronger wage growth under this policy leads to a 0.6 percent increase in consumption. Total hours worked are mostly unaffected under both policies.
The reason for these improvements is twofold. First, both policies replace lower-income immigrants with higher-income immigrants. Higher-income workers have a higher propensity to save, which increases capital, GDP, and wages. Higher incomes also produce more tax revenues, which decreases debt and further increases capital. Second, the increase in the number of high-skilled STEM workers increases TFP, positively affecting GDP, wages, and capital even further.
| 2034 | 2039 | 2044 | 2049 | 2054 | |
|---|---|---|---|---|---|
| Gross domestic product | 0.1 | 0.2 | 0.2 | 0.3 | 0.4 |
| Capital stock | 0.0 | 0.1 | 0.2 | 0.3 | 0.4 |
| Hours worked | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
| Average wage | 0.0 | 0.0 | 0.0 | 0.1 | 0.2 |
| Consumption | 0.1 | 0.2 | 0.2 | 0.3 | 0.4 |
| Debt held by the public | -0.1 | -0.2 | -0.3 | -0.3 | -0.4 |
| 2034 | 2039 | 2044 | 2049 | 2054 | |
|---|---|---|---|---|---|
| Gross domestic product | 0.3 | 0.4 | 0.5 | 0.6 | 0.7 |
| Capital stock | 0.1 | 0.3 | 0.5 | 0.7 | 1.0 |
| Hours worked | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
| Average wage | 0.1 | 0.2 | 0.2 | 0.3 | 0.5 |
| Consumption | 0.3 | 0.4 | 0.4 | 0.5 | 0.6 |
| Debt held by the public | -0.3 | -0.5 | -0.6 | -0.8 | -0.8 |
Source: Penn Wharton Budget Model
Table 3 presents the effects of the two illustrative immigration policies on wages, consumption, and hours worked for each group, including low-skilled, high-skilled non-STEM, and high-skilled STEM workers.
Under the HSI policy, wages for low-skilled workers increase by 1 percent by 2054 (see Table 3.A), while wages for high-skilled workers increase only slightly. Although higher TFP raises wages for all groups, the reduction in low-skilled labor supply drives a more significant wage increase for low-skilled workers because of their increased scarcity and their complementarity with high-skilled labor. In contrast, the increase in high-skilled non-STEM and STEM workers, combined with the reduced availability of low-skilled workers, offsets the TFP-driven wage gains for high-skilled workers.
Table 3.B highlights the stronger wage effects under the HSI STEM policy. With a larger influx of STEM workers and a greater boost to TFP, wage growth is more pronounced across all groups. By 2054, wages for low-skilled workers are projected to be 1.3 percent higher than under current law, while wages for high-skilled non-STEM and high-skilled STEM workers increase by 0.4 percent and 0.3 percent, respectively.
For both policies, changes in consumption closely follow wage trends. Consumption increases among low-skilled workers by between 0.9 and 1.2 percent by 2054, while high-skilled workers’ consumption increases by no more than 0.5 percent. Average hours worked decrease slightly more for low-skilled than high-skilled workers, but those declines are never greater than 0.2 percent.
| 2034 | 2039 | 2044 | 2049 | 2054 | |
|---|---|---|---|---|---|
| Wages | |||||
| Low-skilled | 0.4 | 0.5 | 0.7 | 0.8 | 1.0 |
| High-Skilled non-STEM | 0.0 | 0.0 | 0.0 | 0.0 | 0.1 |
| High-Skilled STEM | 0.0 | 0.0 | 0.0 | 0.0 | 0.1 |
| Consumption per Capita | |||||
| Low-skilled | 0.3 | 0.5 | 0.6 | 0.7 | 0.9 |
| High-Skilled non-STEM | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
| High-Skilled STEM | 0.0 | 0.1 | 0.1 | 0.1 | 0.2 |
| Average Hours Worked | |||||
| Low-skilled | -0.2 | -0.2 | -0.2 | -0.2 | -0.1 |
| High-Skilled non-STEM | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
| High-Skilled STEM | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
| 2034 | 2039 | 2044 | 2049 | 2054 | |
|---|---|---|---|---|---|
| Wages | |||||
| Low-skilled | 0.5 | 0.7 | 0.9 | 1.1 | 1.3 |
| High-Skilled non-STEM | 0.1 | 0.1 | 0.2 | 0.3 | 0.4 |
| High-Skilled STEM | 0.0 | 0.1 | 0.1 | 0.2 | 0.3 |
| Consumption per Capita | |||||
| Low-skilled | 0.4 | 0.6 | 0.8 | 1.0 | 1.2 |
| High-Skilled non-STEM | 0.1 | 0.1 | 0.2 | 0.2 | 0.3 |
| High-Skilled STEM | 0.1 | 0.1 | 0.2 | 0.3 | 0.5 |
| Average Hours Worked | |||||
| Low-skilled | -0.1 | -0.2 | -0.2 | -0.2 | -0.2 |
| High-Skilled non-STEM | 0.0 | 0.0 | 0.0 | -0.1 | -0.1 |
| High-Skilled STEM | -0.1 | -0.1 | -0.1 | -0.2 | -0.2 |
Source: Penn Wharton Budget Model
Table 4 presents the conventional deficit effects of both policies. Outlays decline slightly in both scenarios, primarily due to a reduced number of low-skilled workers relying on income security programs such as Medicaid and SNAP. Over the 2025–2034 period, outlays decrease by $12.3 under the HSI policy and by $13.9 under the HSI STEM policy.
At the same time, revenues increase significantly due to higher tax collections driven by rising incomes. Under the HSI policy, revenues grow by $53.1 billion, while the HSI STEM policy generates an even larger revenue increase of $138.7 billion.
As a result, primary deficits over the 2025-34 period decline by $65.3 billion under the HSI policy and by $152.9 billion under the HSI STEM policy. When accounting for dynamic effects, the reductions amount to $57.0 billion under the HSI policy and $145.2 billion under the HSI STEM policy.
| Provision | 2025 | 2026 | 2027 | 2028 | 2029 | 2030 | 2031 | 2032 | 2033 | 2034 | 2025 - 2034 |
|---|---|---|---|---|---|---|---|---|---|---|---|
| High-Skill Immigration | |||||||||||
| Outlays | -0.3 | -0.5 | -0.7 | -1.0 | -1.1 | -1.4 | -1.5 | -1.7 | -1.9 | -2.0 | -12.3 |
| Revenues | 1.0 | 2.1 | 3.0 | 3.9 | 4.7 | 5.7 | 6.7 | 7.6 | 8.5 | 9.8 | 53.1 |
| Primary Deficit | -1.4 | -2.6 | -3.8 | -4.8 | -5.8 | -7.1 | -8.2 | -9.4 | -10.4 | -11.8 | -65.3 |
| High-Skill STEM Immigration | |||||||||||
| Outlays | -0.4 | -0.6 | -0.9 | -1.1 | -1.3 | -1.6 | -1.7 | -1.9 | -2.1 | -2.3 | -13.9 |
| Revenues | 4.8 | 8.0 | 10.8 | 12.4 | 13.7 | 15.3 | 16.5 | 17.3 | 19.0 | 20.8 | 138.7 |
| Primary Deficit | -5.2 | -8.6 | -11.7 | -13.4 | -15.0 | -16.9 | -18.3 | -19.2 | -21.1 | -23.1 | -152.6 |
| Memorandum: | |||||||||||
| Increase or decrease (-) in primary deficit with dynamic effects, High-Skill Immigration | -2.2 | -3.1 | -4.1 | -4.7 | -5.2 | -5.5 | -3.0 | -9.3 | -9.6 | -10.4 | -57.0 |
| Increase or decrease (-) in primary deficit with dynamic effects, High-Skill STEM Immigration | -6.2 | 5.4 | -13.1 | -15.1 | -15.6 | -17.3 | -18.3 | -19.1 | -22.3 | -23.8 | -145.2 |
Source: Penn Wharton Budget Model
Dynamic distributional analysis considers how a policy affects households across the income and age distribution, including the unborn (represented by a negative age index at the time of the reform). It asks how much, on average, households in each (income, age) bucket value the proposed policy change over their entire lifetime, represented as a one-time transfer at the time of the policy change. Dynamic distributional analysis is the standard in academic research, where conventional analysis is rarely used due to several key limitations that dynamic analysis addresses.
Tables 5.A and 5.B report policy “equivalent variations” for the two policies reported in Table 2 and break down the results by type of worker. A positive equivalent variation means that the person would be better off under the policy reform; a negative equivalent variation means that the person would be worse off under the policy reform. For example, as shown in Table 5.A.A, a low-skilled household aged 60 at the time of the policy change and with a gross income in the highest 20th percent of the income distribution receives $9,665 of value from the HSI policy. Put differently, this household is indifferent between adopting this policy and receiving a one-time payment of $9,665 without this policy change.
While all education, occupation, income, and age groups benefit from the HSI policy, Table 5.A shows that low-skilled workers generally experience greater gains than high-skilled non-STEM workers. With less low-skilled immigration, low-skilled workers’ wages increase more than those of high-skilled workers, contributing to the improvement in welfare. The benefits tend to increase the longer a household is exposed to the policy change, with future generations seeing the largest gains while those nearing retirement benefit the least. Retirees benefit from slightly higher interest rates.
While benefits generally increase by quintile income, low-skilled workers are an exception. Many households in this group rely heavily on means-tested income security programs such as Medicaid or SNAP. Although higher TFP drives wage growth, it can also reduce eligibility for these benefits, partially offsetting the gains from higher earnings.
As Table 5.B illustrates, the HSI STEM policy generally provides greater benefits across all education, occupation, income, and age groups compared to the HSI policy. The benefits that households receive from the HSI STEM policy are, on average, twice as large as those from the HSI policy.
Note: "Gross Income" refers to each household's income in the year of the policy change. We categorize households not yet in the labor force (ages 20 and younger) by their gross income in the year they enter the labor force. We calculate the gross income distribution across all groups of workers. Empty cells in panel A indicate that no low-skilled workers have incomes at that level.
Source: Penn Wharton Budget Model
An overview of our heterogeneous-agent overlapping-generations dynamic model (OLG) is provided in our 2024 publication “Policy Options for Reducing the Federal Debt: Spring, 2024” (available here). Over the next two years, PWBM will release additional details about our model to make our framework more accessible to academics, experts, and policy advocates.
This appendix briefly outlines the new components we have integrated into our model to analyze immigration policies. These enhancements and parameters are either directly sourced from published research or have been newly estimated. The new model components include:
-
Incorporating STEM Workers’ Contribution to Total Factor Productivity (TFP): The model now explicitly accounts for the contributions of both native and immigrant STEM workers to TFP and skill-biased productivity growth. As the STEM workforce expands, technological advancements and process efficiencies accelerate, leading to higher wages, increased investment, and stronger long-term growth.
-
Modeling Skill Complementarities and Substitution Effects: The basic version of our OLG model includes low and high-income workers but treats them as perfect substitutes. To better account for the different effects that natives and immigrants have on labor supply and the economy, we introduce a framework that differentiates between low-skilled, high-skilled non-STEM, and high-skill STEM workers.
The following empirical concepts are now included in our analysis:
-
Immigration projections: The PWBM produces immigration projections that we now use in the OLG model, including immigrants’ education and occupation groups.
-
Estimate of the STEM workers’ contribution to Total Factor Productivity: The PWBM now incorporates empirical evidence on the relationship between TFP and the number of workers in STEM occupations. Through a linear regression analysis, we estimate a constant elasticity of TFP with respect to the number of STEM workers using data on labor supply from the Current Population Survey’s Annual Social and Economic Supplements (ASEC), estimates of the capital stock from the Bureau of Labor Statistics (BLS), and gross domestic product data from the National Income and Product Accounts (NIPA).
-
Parameters determining the degree of substitutability or complementarity between groups: We use estimated parameter values from economic literature to assess how low-skilled, high-skilled non-STEM, and high-skilled STEM workers interact in the labor market.
-
Estimates of group-specific earnings processes: Members of the different skill and occupational groups face different earnings processes, which we estimated from the Panel Study of Income Dynamics (PSID) and the CPS ASEC.
Felipe Ruiz Mazin and Felix Reichling produced this analysis under the direction of Felix Reichling and Kent Smetters. Mariko Paulson prepared the brief for the website.
This material is based upon work supported by Alfred P. Sloan Foundation under Grant No. G-2024-22429.